Circulation and suspended sediment transport in a coral reef ... - LEGOS

Marine Pollution Bulletin 61 (2010) 269–296

Contents lists available at ScienceDirect

Marine Pollution Bulletin journal homepage: www.elsevier.com/locate/marpolbul

Circulation and suspended sediment transport in a coral reef lagoon: The south-west lagoon of New Caledonia S. Ouillon a,b,*, P. Douillet c, J.P. Lefebvre c, R. Le Gendre d, A. Jouon e, P. Bonneton e, J.M. Fernandez d, C. Chevillon d, O. Magand f, J. Lefèvre d, P. Le Hir g, R. Laganier h, F. Dumas g, P. Marchesiello a,d, A. Bel Madani i, S. Andréfouët d, J.Y. Panché d, R. Fichez c a

Université de Toulouse, UPS (OMP-PCA), LEGOS, 14 av. Edouard Belin, Toulouse F-31400, France IRD, LEGOS, 14 av. Edouard Belin, Toulouse F-31400, France IRD, UAM, Dpt Hidrobio – DIV de Ciencas Bio y Salud, AV. San Rafael Atlixco 186 – Col. Vicentina 09340, Mexico DF, Mexico d IRD Nouméa, BP A5, 98848 Nouméa cedex, New Caledonia e Université de Bordeaux, CNRS, UMR 5805, Bordeaux F-33000, France f LGGE, CNRS, Université Joseph Fourier, 54 rue Molière, Saint Martin d’Hères cedex F-38402, France g DYNECO/PHYSED, Ifremer, BP 70, Plouzané F-29280, France h Université Paris Diderot – Paris 7, Case 7001, 75205 Paris cedex 13, France i IRD, IMARPE/CIMOBP, Instituto del Mar del Perù, Esquina de Gamarra y General Valle S/N Chucuito, Callao, Peru b c

a r t i c l e Keywords: Hydrodynamics Suspended sediment New Caledonia Coral reef lagoon Sediment Turbidity Resuspension

i n f o

a b s t r a c t The south-west lagoon of New Caledonia is a wide semi-open coral reef lagoon bounded by an intertidal barrier reef and bisected by numerous deep inlets. This paper synthesizes findings from the 2000–2008 French National Program EC2CO-PNEC relative to the circulation and the transport of suspended particles in this lagoon. Numerical model development (hydrodynamic, fine suspended sediment transport, windwave, small-scale atmospheric circulation) allowed the determination of circulation patterns in the lagoon and the charting of residence time, the later of which has been recently used in a series of ecological studies. Topical studies based on field measurements permitted the parameterisation of wave set-up induced by the swell breaking on the reef barrier and the validation of a wind-wave model in a fetch-limited environment. The analysis of spatial and temporal variability of suspended matter concentration over short and long time-scales, the measurement of grain size distribution and the density of suspended matter (1.27 kg l1), and the estimation of erodibility of heterogeneous (sand/mud, terrigenous/biogenic) soft bottoms was also conducted. Aggregates were shown to be more abundant near or around reefs and a possible biological influence on this aggregation is discussed. Optical measurements enabled the quantification of suspended matter either in situ (monochromatic measurements) or remotely (surface spectral measurements and satellite observations) and provided indirect calibration and validation of a suspended sediment transport model. The processes that warrant further investigation in order to improve our knowledge of circulation and suspended sediment transport in the New Caledonia lagoon as well as in other coral reef areas are discussed, as are the relevance and reliability of the numerical models for this endeavour. Ó 2010 Elsevier Ltd. All rights reserved.

1. Introduction * Corresponding author at: Université de Toulouse, UPS (OMP-PCA), LEGOS, 14 av. Edouard Belin, Toulouse F-31400, France. Tel.: +33 5 61 33 29 02. E-mail addresses: [email protected] (S. Ouillon), [email protected] (P. Douillet), [email protected] (J.P. Lefebvre), [email protected] (R. Le Gendre), [email protected] (A. Jouon), [email protected] (P. Bonneton), [email protected] (J.M. Fernandez), [email protected] (C. Chevillon), [email protected] (O. Magand), [email protected] (J. Lefèvre), [email protected] (P. Le Hir), [email protected] (R. Laganier), [email protected] (F. Dumas), [email protected] (P. Marchesiello), [email protected] (A. Bel Madani), [email protected] (S. Andréfouët), [email protected] (J.Y. Panché), renaud.fi[email protected] (R. Fichez). 0025-326X/$ - see front matter Ó 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.marpolbul.2010.06.023

1.1. General A main interest (and also difficulty) of scientific research in coastal waters is due to the overlap of large variety of biotic and abiotic processes, the biological richness of these systems, the socio-economic importance of these areas and the interrelationship between all these factors. Particle dynamics is a topic which illustrates this complexity. The south-west lagoon of New Caledonia is of particular interest because it is typical example of a South Pacific barrier reef lagoon. In addition, sediment transport is intrinsically

270

S. Ouillon et al. / Marine Pollution Bulletin 61 (2010) 269–296

linked with metal contamination resulting from industrial and urban waste, coastal management, changes in land use and, above all, mining activities, all of which have an impact on the lagoon ecosystem. New Caledonia is currently the third largest producer of nickel in the world. The saprolitic layers and enriched laterites have been exploited for more than a century using the opencast mining technique. Moreover, the prospecting and open cast mining on the hilltops has resulted in major alterations to landscape and habitat strongly enhancing natural soil erosion and subsequent river transport of terrigenous material and associated metals toward the lagoon (Bird et al., 1984). Although all pelagic and benthic communities are affected to some extent by the increases in sediment deposition and decreases in light penetration, fringing reefs are especially sensitive to these changes. This paper synthesizes findings from the 2000–2008 French ‘‘Programme National Environnement Côtier” (PNEC) relative to the circulation and the transport of suspended particles in the south-west lagoon of New Caledonia. The questions asked in this project are detailed in Section 1.2. Section 2 presents the study area, its geomorphology, climate, temperature and salinity climatology, tide and wave regimes. Hydrodynamic results (circulation, residence time, exchanges with the ocean, wind waves, atmospheric forcing) are presented and discussed in Section 3. After a detailed presentation of the lagoon sedimentology, Section 4 deals with suspended sediment transport. This section analyzes the spatial and temporal variability of turbidity, the grain size distribution of suspended particles, the optical measurements, modelling of sediment transport, erodibility of sediment beds, and ends with a sedimentary budget. 1.2. Historic studies and the PNEC program The Institute of Research for Development (IRD, formerly called ORSTOM) initiated a pluridisciplinary scientific program (‘‘ECOTROPE”) in 1996 to study the impact of anthropogenic activities on the functioning of the south-west lagoon of New Caledonia (e.g. Bujan et al., 2000). Based on ECOTROPE, the PNEC funded a large interdisciplinary study where scientists from the IRD and other French scientific institutions such as the universities and CNRS worked together to improve our knowledge of its physical, chemical, biological functioning in relation to its use. Physical processes in coral reef lagoons have already been individually studied in detail and the variability of some physical parameters have been synthesized. Hydrodynamics has been widely studied (Wolanski and Pickard, 1985; Hearn and Parker, 1988; Atkinson et al., 1992), occasionally on the basis of numerical models (Smith, 1985; Frith and Mason, 1986; Prager, 1991; Tartinville et al., 1997; Kraines et al., 1998, 1999; Douillet et al., 2001). Wave-associated processes in lagoons have been investigated by, e.g. Hardy (1993), Symonds et al. (1995), Hearn (1999), Callaghan et al. (2006). The impact of hydrodynamics on lagoon functioning was analyzed by Bradbury and Young (1981), Hamner and Wolanski (1988), Roberts et al. (1992), Wolanski (2001). Studies have also dedicated to variability in temperature (Liston et al., 1992; Nadaoka et al., 2001a; Ouillon et al., 2005) and in salinity (Ouillon et al., 2005) in lagoons. Among coastal systems, coral reefs are thought to be particularly vulnerable to the amount of sediment provided by anthropogenic activities (Smith and Buddenmeier, 1992). As sedimentation is a potential factor driving reef degradation (Rogers, 1990), sediment transport and fluxes have been monitored and analyzed over coral reef lagoons (Hamilton, 1994; Larcombe et al., 1995, 2001; Orpin et al., 1999; Kraines et al., 2001), with some studies describing and quantifying the impact of storms and hurricanes on sediments (Bonem, 1988; Gagan et al., 1990). Although the variability of suspended matter concen-

trations at different time scales has been analyzed in continental shelves and bays (Castaing et al., 1999; Schoellhamer, 2002), few authors have conducted such studies in coral reef lagoons (Thomas et al., 2003). Even if physical processes in reef lagoons have been investigated, few studies have yet presented integrated physical functioning of coral reefs over areas of great extent, with the exception of the Great Barrier Reef (Wolanski, 1994). Macroscale processes influence large portions of a continental shelf, mesoscale processes determine the degree to which individual reef systems are connected to their oceanic environment, and microscale processes affect zones within reefs (Andrews and Pickard, 1990). Physical analysis integrating temperature, salinity, currents and waves have been presented by Wing and Leichter (1997) and Nadaoka et al. (2001b). However, it is clear that additional integrated physical studies of reefs at different scales and over extended periods are needed. This paper is an attempt to summarize and synthesize the physical studies (circulation and suspended sediment transport) which were performed in the south-west lagoon of New Caledonia. This program follows on from research funded by the IRD and from the French National Program on Coral Reefs, led by Jacques Clavier and Renaud Fichez. During these two programs, a large number of hydrological and current measurements were made (Douillet et al., 1989, 1990), enabling the set up a 2D model of tide propagation (Douillet, 1998) followed by the development of a 3D model of hydrodynamics in collaboration with Ifremer. At the same time, the leading paper by Hearn (2001) of the special issue of Coral Reefs devoted to ‘‘Coral reef hydrodynamics” highlighted the context and importance of such studies. Hearn explained that the 9th International Coral Reef Symposium in Bali in 2000 saw for the first time a session devoted entirely to hydrodynamics that ‘‘in a sense, marked the coming of age of ‘coral reef hydrodynamics’ ”. Obviously, prior to this conference, scientists had not been ignoring the study of lagoon hydrodynamics, particularly given the nuclear testing that was carried out by the United States and France in some Pacific atolls (e.g. von Arx, 1948). Andrews and Pickard (1990) wrote a brilliant synthesis of this pioneer work on lagoon hydrodynamics. However, the massive coral bleaching episodes in the 1980’s and that of the El Niño event of 1997–98 brought renewed interest in the physics, biology, chemistry and ecology of coral reef environments. The physics-focused paper published in the special issue of Coral Reefs, Douillet et al. (2001), presented the first application of a 3D model of fine suspended sediment transport in the south-west lagoon of New Caledonia and posed the central questions relevant to this review: (1) What kind of easy-to-use parameters can we propose for characterization of hydrodynamics for chemical or biological applications? (2) Is it possible to estimate the fluxes of water and momentum above the barrier reef induced by the swell surge? (3) Is it possible to estimate the wind wave action on resuspension within the lagoon? (4) How can we take into account a realistic wind stress field rather than a uniform stress in the circulation model (in relation with orographic effects)? (5) What is the size distribution of in-water particles and their density? (6) What is the sediment erodibility and how can it be connected to sedimentological parameters? These questions were approached progressively during the PNEC program from 2000 to 2008. The answers are synthesized in the present paper. As always in research, the initial questions


have evolved and some new directions that may help structure of future research are given at the end of the paper. 2. The study area 2.1. Geomorphology New Caledonia is located 1500 km east of Australia, at the southern end of the Melanesian Arc near the Tropic of Capricorn (Fig. 1). The main island is surrounded by barrier reefs that delimit large lagoon areas. The reefs of New Caledonia cover a total area of 4537 km2 and the non reef areas (lagoons, terraces, enclosed basins and inlets) 31, 336 km2 (Andréfouët et al., 2009). Most of the population lives along the shore of the south-west lagoon where Nouméa, the main city, is located. Geomorphologically, the lagoon is funnel shaped, with the coast on the eastern side and the barrier reef on the west side. The distance from the coast to the barrier ranges from 5 km (at North) to 40 km (at South; see Fig. 1) and the length along the North-West–South-East axis is about 100 km. Deep channels cut the barrier reef with the crest emergent during the lowest tides. This area is called the southwest lagoon of New Caledonia (hereafter denoted SLNC). Its mean water depth is 17.5 m, but maximum depths of about 60 m are observed in narrow canyons located in the channels. The SLNC is open to the southern lagoon at its southeastern end where it receives the main input of oceanic waters. The tides are mainly semi-diurnal with a tidal range varying from about 1.4 m at spring to 0.6 m at neap (Douillet, 1998). Three geomorphological zones can be distinguished: the bays or inner lagoon, the outer lagoon, and the inner reef (Dandonneau et al., 1981). The current morphology of the SLNC results from the last great variation of sea level which occurred during the Würmien Ice Age (18,000 years BP). When sea level was about 120 m below the current zero, valleys, canyons and channels in the lagoon and in the reef barrier were created (Coudray, 1975). In the south-western part of the mainland of New Caledonia, soils are composed of peridotites, harzburgites with orthopyroxenes composed of magnesium and iron silicates, and infiltrated by transition metals like Ni, Co, Cr, Fe and Mn (Lillie and Brothers, 1970; Paris, 1981). The weathering of peridotites results in the formation of nickel- and iron-rich smectites, serpentine, goethite and hematite (Trescases, 1973), and in the migration of transition met-

271

als. The latter are strongly concentrated along the weathered profile in the saprolite (also called ‘‘garnierite”) and laterite layers which are subject to mining extraction (Fernandez et al., 2006). The extensive extraction method requires, initially, the elimination of vegetation, the extraction of the ferralitic layer, followed by the laterites of low content before reaching the exploitable saprolites. This method produces significant volumes of deads (approximately 2.5 times the volume of actually exploited ore), that are generally left next to the extraction site or are contained by filtering dams of uncertain efficiency (Fernandez et al., 2006). It is estimated that a total of 300 M m3 of soil has been displaced by direct extraction since the beginning of mining activities (Labrosse et al., 2000). 2.2. Climate, rainfall and river discharge Climate in New Caledonia is mainly characterized by a dry-tropical regime with alternation of dry and wet seasons. The system is strongly influenced by the El Niño Southern Oscillation (ENSO), has south-easterly trade winds, irregular rainfall, episodic cyclones, long swell coming irregularly from the South and South-East, and torrential hydrological regimes. Climate is mainly constrained by the position of the Intertropical Convergence Zone (ITCZ) and South Pacific Convergence Zone (SPCZ) (Pesin et al., 1995). From January to March (warm season), the ITCZ is in its southernmost position, rainfall is heavy and tropical storms or cyclones can affect New Caledonia. When the ITCZ moves to the northern hemisphere (mid-May to mid-September) cold fronts carrying polar air from the Tasman Sea can occasionally bring heavy rainfall. The intermediate periods (April–May and September–December) are characterized by sunny and dry weather. Mean seasonal variations of wind regime are characterized by two major seasons (Pesin et al., 1995). From October to May, south-easterly (110°) trade winds dominate with a mean velocity of 7 m s1. From June to September, wind velocity is weaker and direction is more variable. Inter-annual variations elicit stronger (weaker) winds during El Niño (La Niña) (Ouillon et al., 2005). Average freshwater inputs to the lagoon are weak since the catchment areas are small and, when freshwater inputs occur, they are generally reduced to a period of a few days at most. The main freshwater inputs to the SLNC come principally from the Dumbéa River (drainage area: 238 km2), La Coulée River (85 km2) and Les Pirogues River (140 km2) (Fig. 1). Over the 1997–2001 period,

Fig. 1. Map of the south-west lagoon of New Caledonia. Blue color corresponds to the water depth (scale to the right, in meter). The red circle indicates the weather station on the Maître Island reef. The purple star indicates the Amédée Lighthouse. The black box on the Aboré reef shows the area where fluxes were quantified over the reef (Section 3.3). Wave measuring locations are shown by yellow diamonds (Section 3.4).

272


mean flows were 11.5 m3 s1 for the Dumbéa River, 3.5 m3 s1 for the Coulée River and 5.2 m3 s1 for the Pirogues River; mean monthly discharges are shown on Fig. 2. The 5-year period is too short to estimate long-term values because of the high inter-annual variability of rainfall and river discharge and because it included both one El Niño and one La Niña events (Ouillon et al., 2005). However, the order of magnitude of river discharge can be compared to the flow of ocean water entering and leaving the SLNC resulting from tidal currents and trade wind-induced circulation. Ocean water flow is generally on the order of 80,000 m3 s1 regardless of water inputs over the reef (Jouon et al., 2006) and so the lagoon waters are therefore mainly of oceanic origin. Rainfall and river discharge vary from one year to another significantly. Nicet and Delcroix (2000) analyzed the correlation between anomalies in precipitation or river discharge in New Caledonia with the Southern Oscillation Index (SOI), which is a good indicator of ENSO variations. They showed that: (1) ENSO modulates rainfall regime in New Caledonia with a 20–50% decrease (increase) during El Niño (La Niña) events; (2) the flow rate of two rivers at the north end of the SLNC can almost double during La Niña and decrease by more than 50% during El Niño; (3) the best lagged correlations are obtained when SOI leads rainfall by 2– 3 months and river flow rates by 3–6 months. The major rivers of the south-western part of New Caledonia are characterized by an upper course with abrupt relief and a lower course made up of piedmont deposits and bedrock weathering material. Due to the torrential regime, sediment supply is very low in the absence of rainfall because of the low flow energy (Baltzer and Trescases, 1971b; Dugas and Debenay, 1982). On the contrary, as rainfall can reach up to 700 mm in 24 h, the induced intense weathering results in extremely high quantities of suspended matter. In one 24 h period, more than 20,000 tonnes of solid matter were delivered by the Dumbéa River during cyclone Brenda in 1968 (Baltzer and Trescases, 1971a). These particles strongly colour the river waters and plume and can be visible by spatial observations.

perature) outside the lagoon, which were described by Gouriou and Delcroix (2002). This analysis was proposed by Ouillon et al. (2005) based on 20 field surveys spanning a 4-year period (1997–2001). They concluded that ENSO related regional SSS anomalies (0.2) were twice as large as the seasonal variations and that this brought less saline waters into the lagoon during a La Niña event. The SSS in coastal areas was also influenced by the meteorological and hydrological consequences of ENSO with salinity differences within the lagoon ( 2 °C in December 1997; Ouillon et al., 2005). Thermal stratification sets up during the warm and dry season from October to December and a thermo-haline stratification is observed during the warm and wet season from January to April. The existence of a haline stratification all the year long depends on the river discharges and is thus more likely during La Niña periods. Thermo-haline stratification is thus enhanced in summer during La Niña periods. Haline stratification, if any, is enhanced in coastal areas. Accordingly, the bays and coastal zones amplify seasonal and inter-annual regional salinity variations due to ENSO and amplify the regional variations of temperature leading to higher crossshore thermal gradients in summer during El Niño.

2.3. Temperature and salinity To understand the main variations in temperature and salinity within the lagoon under non-storm conditions, it is necessary to compare their distributions against seasonal and ENSO-related variations of oceanic SSS (Sea Surface Salinity) and SST (Sea Surface Tem-

45.

Mean monthly discharge (m 3 s-1)

40.

Pirogues R. Coulée R.

30.

Dumbéa R.

20.

10.

0. ry

y

ar

u an

J

ua

F

r eb

ch

M

ar

ril

Ap

ay

M

ne

Ju

ly

Ju

r

t Se

p

m te

r

er

be

us

g Au

O

r

be

ob

ct

N

o

m ve

be

m ce

e

D

Fig. 2. Mean monthly discharge of the Pirogues, Coulée and Dumbéa Rivers calculated for the period 1997–2001.


2.4. Tide and wave regimes The harmonic analysis of tidal gauge measurements showed that M2 is the dominant tidal signal in the SLNC (roughly 3 K1, 5 O1, 9 P1, 5 N2, 3 S2) (Douillet, 1998). Using the amplitude ratio (O1 + K1)/(M2 + S2), the tide is determined as mixed and mainly semi-diurnal. Tidal streams are mostly generated by the M2 constituent. Amplitude and phase of M2 increases regularly from the South to the North (Douillet, 1998). Mean tidal current velocity is about 0.05–0.1 m s1 inside the lagoon and about 0.2– 0.3 m s1 in the passages (Jarrige et al., 1975; Douillet et al., 1989). M2 currents are maximum in the South and in the passages. Currents generated by S2 are one third of M2, and those by K1 are one tenth of M2 (0.01 m s1), except at the inlets. S2 amplitude increases from South to North more slowly than M2. S2 propagates from North-West to South-East. S2 currents are around 0.03 m s1 over most of the lagoon but can locally reach 0.1 m s1 at the inlets. Waves in the lagoon have two origins: the oceanic swell that penetrates into the lagoon and the wind-waves that develop within the lagoon. Oceanic swell breaks along the barrier reef and penetrates through the passages. Swell breaking induces fluxes of water and momentum to the lagoon (see Section 3.3). To our knowledge, no specific review of wave climate around New Caledonia has been performed. However some parameters can be estimated from the available world wave climatologies obtained from a combination of satellite remote sensing and model predictions (e.g. Young, 1999) or from model reanalysis (Sterl and Caires, 2005; Caires et al., 2005). The mean yearly significant wave height is approximately 2 m (Young, 1999) or 2.1 m (Caires et al., 2005). Three periods can be roughly distinguished from Caires et al. (2005): the highest swell comes from the South–South-East between March and May (mean monthly values range: 2.3–2.4 m), high swell also occurs from June to September and comes from the South (2.1–2.2 m), and the lowest swell comes from the South-East between October and January (1.9–2 m). Estimations of the mean yearly wave period around the southern part of New Caledonia ranges from 7.1 s (Caires et al., 2005) to 8.7 s (Young, 1999), with slightly longer periods from March to October than from November to February but the differences are on average low (around 0.4 s between the lowest and the highest mean monthly wave periods). The influence of swell inside the lagoon strongly depends on its direction. Short episodes of westerly swell bring high energy up to the shore of the SLNC. The impact of other swell conditions is generally less important and strongly depends on the vicinity to the inlets and, sometimes, to the barrier reef (see Section 3.3). Wind waves quickly develop and propagate in the lagoon. No wind wave climatology in the lagoon of New Caledonia is yet available in the literature. The highest wind waves in the outer lagoon occur when the fetch is the longest, i.e. during south-easterly trade wind episodes. During these episodes, the significant wave height and period in the SLNC are typically between 0.8 and 1.4 m and around 3 s, respectively (Jouon et al., 2009 – see also Section 3.4). The highest wind waves near shore are observed during the short and highly energetic episodes of westerly wind and, obviously, during cyclones.

3. Hydrodynamic characterization 3.1. Circulation Hydrodynamics in the SLNC has been studied for 30 years. The first analysis showed that wind and tide were the main forces generating circulation (Jarrige et al., 1975; Morlière and Crémoux,

273

1981; Morlière, 1985; Rougerie, 1985). Intensive measurements of tide, wind and currents were performed at the end of the 1980’s (Douillet et al., 1989, 1990). During the 1990’s, a 3D hydrodynamic model precursor of MARS3D developed by IFREMER (Lazure and Salomon, 1991; Lazure and Dumas, 2008) was adapted and validated for the SLNC (Douillet, 1998). The governing equations of the hydrodynamic model express the conservation of mass, momentum and salinity, and assume the hydrostatic and Boussinesq approximations. The equations are solved in the r-coordinate system using a finite difference technique and the resolution is based on the mode splitting technique (Blumberg and Mellor, 1987). The version used at the beginning of the PNEC program involved a mixing length turbulence model (Douillet et al., 2001). The mesh grid extended from the south of the main island to the north of Nouméa under an alongshore distance of 150 km, the horizontal grid size was 500 m and each water column was divided into 10 r-levels (Douillet, 1998; Douillet et al., 2001). This model was one of the first 3D models applied to hydrodynamics in coral reef environments after Tartinville et al. (1997). This 3D hydrodynamic model was also used to calculate the fluxes within the lagoon in a 3D biogeochemical model (Pinazo et al., 2004). The semi-diurnal tide propagates from the South to the North (Douillet, 1998). Local south-easterly trade winds drive a general North-West drift (Douillet et al., 2001). Response of the dynamics to the wind action is fast. In the absence of local wind, velocity at 20 m depth shows two peaks per tidal cycle during flood and ebb. During trade wind episodes, velocity direction in the lagoon remains in the same quadrant and presents only one significant peak during flood. Trade winds generate a strong sub-surface gyre north of Ouen Island (Fig. 3b). Blowing from the south-east, wind drives the surface water to the North-West which creates a gradient in the sea-surface. A sub-surface return current develops along the coast to equilibrate the pressure gradient and feeds the subsurface gyre (Douillet et al., 2001). The long term tidal water displacement is well described by the mean lagrangian residual velocity. Its value for the M2 constituent (Fig. 3a) can locally reach 0.03 m s1, but is generally close to 0.01 m s1 (Douillet, 1998). Water mainly enters the SLNC at the South between Ouen Island and the barrier reef. One part flows directly to the passage while the other part flows northwards and leaves the lagoon by the Dumbéa pass. A cape effect generates a vortex on the West side of Ouen Island and near shore, residual currents are weak. Considering long term transport, the lagrangian residual velocity generated by the tidal circulation is one order of magnitude lower than the mean vertical current due to trade winds (Fig. 3 a and b). Although instantaneous tide-induced and wind-induced currents are of the same order, the tidal lagrangian residual velocity is around 1 cm s1 whereas wind-induced velocity is 10 cm s1 or higher. Tide- and wind-induced circulation flushes the lagoon which empties through the inlets. Over the reef, the flow may alternatively enter and leave the lagoon during trade wind episodes. Under West wind or weak wind, when the oceanic swell is high, flow can be positive towards the lagoon above the reef regardless of the tidal period (see Section 3.3). This flow and the corresponding setup generate a current along the internal reef which flushes the white sand areas and leaves the lagoon through the inlets. In 2007–2008, an exercise of model validation was performed using drifters (Niiler et al., 1995). Monitoring of water layers by means of Argos drifters attached to a weighted drogue (‘holey sock’) is mainly used in coastal oceanography to investigate rip currents (e.g. Kennedy and Thomas, 2004; Reniers et al., 2007). The thickness of layers to be monitored was fixed to 5 m and the 15 m sock was cut accordingly. Tests have been conducted to estimate the vertical shear and the variation of vertical position

274


0.05 m/s

a

22°10'S

22°30'S

166°10'E

166°30'E

166°50'E

0.5 m/s 22°10'S

b

22°30'S

166°10'E

166°30'E

166°50'E

Fig. 3. Circulation in the south-west lagoon of New Caledonia: (a) Lagrangian residual mean velocity of the M2 tidal constituent; (b) depth-average current generated by trade wind (110°, 8 m s1) with no tide.

around the nominal depth of the sock. A pressure gauge (precision pressure sensor, Drück) and acoustic Doppler velocimeter (ADV, Sontek) have been alternatively fixed to the top and at the bottom of the drifter. The results showed that, apart from when the drifter contacted the seabed as detected by the velocity deduced from Argos positions, the slip and variation of height in the water column of the drifters were negligible, compared to mean monitored velocity and wave heights. Comparisons were performed between measurements and simulations for the trajectories and velocities and an example is given in Fig. 4. The analysis of combined drifter experiments and numerical simulations in the SLNC is in progress. 3.2. Residence time The immediate products of a 3D hydrodynamic model such as current fields or fluxes between boxes are not always useful for

chemical, geochemical, biological or ecological applications where only simple indicators of dynamics are required. These data are often a consequence of the transformation processes, chemical reactions, and mechanisms of exchange between benthos and pelagos or between dissolved and particulate elements that are applied to the water mass after its entrance into the lagoon. It is therefore interesting to produce hydrodynamic models that provide responses on similar timescales. This question was examined in the SLNC within the multi-disciplinary context of the PNEC. The literature on hydrodynamic time parameters refers to different variables commonly called ‘‘residence time”. The paper by Jouon et al. (2006) proposed a synthesis of these global and local parameters, calculated them in the SLNC considering a simplified tide represented by M2 and S2 and mean constant trade wind, then compared the results. On the global scale, the water exchange time, the average export time and the


275

Fig. 4. Comparison between a measured drifter trajectory (in blue) and computed trajectories obtained from lagrangian simulations using MARS3D. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

a

21°50'S

n ai M nd La

22°10'S

r Ba

D B

f ee

rR

rie

22°30'S

110.0 90.0 70.0 50.0 40.0 35.0 30.0 25.0 20.0 15.0 10.0 9.0 8.0 7.0 6.0 5.0 4.0 3.0 2.0 1.0 0.5

C A

Days

165°50'E

166°10'E

166°30'E

166°50'E

b

21°50'S

n ai M nd La

22°10'S

e rri Ba f ee rR

22°30'S

110.0 90.0 70.0 50.0 40.0 35.0 30.0 25.0 20.0 15.0 10.0 9.0 8.0 7.0 6.0 5.0 4.0 3.0 2.0 1.0 0.5

Days

165°50'E

166°10'E

166°30'E

166°50'E

Fig. 5. Local e-flushing time (a) and flushing lag (b) within the SLNC (in days). Four main water types are distinguished from their e-flushing time. A: water rapidly renewed (25 days).

e-flushing time (as defined hereafter, in concordance with Jouon et al., 2006) were calculated over three embedded control volumes. The export time, flushing lag and local e-flushing time were also mapped over the SLNC. Water exchange time is defined as the ratio between the total volume and the daily volume flux entering or leaving the domain. Its value is 6.8 days in the SLNC. Export time is locally defined as the time needed for a water particle to leave the lagoon, and is calculated by lagrangian tracers. Its average value over the calculation domain is called the average water export time (10.8 days over the SLNC). At an initial time, the water mass in the domain is uniformly ‘‘colored” by a virtual tracer at concentration C0, then every new water parcel entering the domain is associated to a nil concentration of that tracer. The time at which the average tracer concentration is reduced by a factor 1/e is called the e-flushing time. It takes in average 12.5 days for 65% (1–1/e) of initial water to leave the lagoon. Locally, the time needed for the concentration to be reduced by a factor 1/e was called the local e-flushing time, and the one required for the concentration to be reduced by 5% (i.e. the time required for a significant amount of renewed water to reach the considered point) is called the flushing lag (Fig. 5b). The flushing lag relates to the minimum age of water mass. Its addition to the export time gives a low estimate of the transit time. The map of local e-flushing time (Fig. 5a) shows how the residence time is very heterogeneous within the lagoon, from very short values (less than one day around channels) up to two months at the head of bays. The gyre extending between the North-West of the Ouen Island and the ‘‘banc Gail” area is characterized by long eflushing time of around one month. Excepting this characteristic, the e-flushing time distribution shows four types of waters which can be distinguished from their history (Fig. 5a): water renewed very quickly, with e-flushing time less than two days and located at the South of a line running from the U Reef at the South to the Uitoé pass at the North (type A); a littoral ‘‘river” flowing alongshore, with residence time of 9–25 days, draining the waters that enter the lagoon through the Woodin channel (type B); intermediate waters between them, with residence time of 2–9 days, which result from the mixing of type A and type B waters by the tidal oscillation (type C); and old waters with e-flushing time higher than 25 days at the head of bays (type D).

276


Amongst global residence times, the average export time was shown to be nearly 0.7 times the e-flushing time in the case where concentration decrease is exponential, while the comparison between average export time and exchange time were shown to provide information on the relative effectiveness of export and exchange processes, i.e. to distinguish short transit areas and retention zones. Local residence times in a calculation domain provide more refined information about mixing, and enable the identification of areas of particular hydrodynamic regime. Calculations in the whole SLNC and in two sub-domains showed a higher sensitivity in the actual values of residence times than in the general patterns of their spatial distributions. The study by Jouon et al. (2006) had some limitations: calculations were made considering uniform and constant wind conditions, and without the exchange of breaking oceanic swell on the reef. It is now possible to estimate the influence of waves breaking over the reef (Section 3.3) and to take into account an actual wind distribution (Section 3.5). However, taking into account true nonstationary meteorological conditions may be problematic. Despite these limits, the paper by Jouon et al. (2006) provided relevant information for the most frequent case of trade wind. Moreover, these values have recently been used to understand the dynamics of biological productivity (phytoplankton: Torréton et al., 2007; organic matter particle dynamics: Mari et al., 2007), residence time and dissolved metal concentrations (Migon et al., 2007), and as an indicator of anthropogenic influence on water quality around Noumea, in relation to the fish populations (Sasal et al., 2007).

Fig. 7. Measured cross-reef current ur on the reef-top (top, yellow line) and comparisons between the mean water depth at the reef-top (with no setup), dr (blue bold line) and the water height at the breaking point, hb (red thin line), during the experiment. In abscissa, numbers refer to successive days. The points where ur is maximum are indicated in red on top and with arrows below. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

3.3. Flux over the barrier reef As waves break on a reef, they create a radiation stress gradient that drives wave-setup and wave-induced currents. These mechanisms which are typical of coral reef environments exert a major influence on the hydrodynamics, morphology and biological variability of shallow submerged reefs. Since the set-up can achieve 20% of the swell height (Tait, 1972; Gourlay, 1996), these processes

Fig. 6. Measured wave setup fr on the reef-top (top, green line) and comparisons between the mean water depth at the reef-top (with no setup), dr (blue bold line) and the water height at the breaking point, hb (red thin line), during the experiment. In abscissa, numbers refer to successive days. The points where fr is maximum are indicated in red on top and with arrows below. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

may significantly alter the circulation and flushing of lagoons. The tidal evolution of water depth over the reef top strongly influences wave-induced phenomena (Gourlay and Colleter, 2005). However, only a few field experiments have been devoted to the analysis of both tidally modulated wave-induced current (e.g. Tartinville and Rancher, 2000) and tidally modulated wave-setup as the deployment of instruments on a reef is not simple. In the hydrodynamic model presented by Douillet et al. (2001), the exchanges between the ocean and the lagoon over the barrier reef were only schematic. Therefore, it was important to characterize wave-setup and waveinduced flows over the New Caledonia reef. A field experiment was conducted on the Aboré coral reef in October 2005 in order to measure the wave induced flow over the reef. A series of sensors were deployed along a reef transect to obtain measurements of the cross-shore propagation of the waves. Measurements were made of the tidal level evolution and incident wave characteristics on the external reef slope by a nondirectional Wave and Tide Recorder (WTR9, Aanderaa). Velocity and pressure on the reef flat were measured using two acoustic Doppler velocimeters (ADV Vector, Nortek), each of them being coupled with two pressure sensors, and the current at the inner end of the transect was measured by a S4 InterOcean electromagnetic current meter (see the cross-shore profile in Fig. 2, Bonneton et al., 2007). The bathymetry of the study area was measured using a DGPS (Trimble 5700). Pressure measurements, corrected for atmospheric variations, were converted to water depth, and by taking in account topography elevation, to sea surface elevation f. On the reef flat, the ratio of water depth to wavelength was very small; consistent with the shallow-water approximation, f was estimated assuming a hydrostatic pressure distribution. Bonneton et al. (2007) present the first analysis of these data, focused on the first sea-state regime (from days 0 to 7) when cross-reef current ur was mainly positive (i.e. lagoonward) and reached a maximum value of 0.65 m s1. They show that tidally modulated wave setup and wave-induced current on the Aboré reef are mainly controlled by the difference between depth at

277


>

25.0

22.5 - 25.0

21°50'S

20.0 - 22.5 17.5 - 20.0 15.0 - 17.5 12.5 - 15.0 10.0 - 12.5

n ai M

7.5 - 10.0

nd La

22°10'S

5.0 -

7.5

2.5 -

5.0

0.0 -

2.5

-2.5 -

0.0

-5.0 - -2.5

er rri Ba

-7.5 - -5.0

f ee R

-10.0 - -7.5 -12.5 - -10.0 -15.0 - -12.5 -17.5 - -15.0

22°30'S

-20.0 - -17.5 -22.5 - -20.0 -25.0 - -22.5
hb/2 (during the days 7–12, not shown on Fig. 7), there was only one maximum of ur per tidal cycle when dr was minimum, in concordance with Eq. (2), i.e. at low tide, 180° out of phase with dr. These new results mean that reef fluxes can now be taken in account in models of lagoon hydrodynamics. This is particularly important where significant exchanges induced by the breaking of oceanic swell occur. The deriving flow is probably responsible for the flushing of the ‘‘inner slope hydraulic sand bank” (Battistini et al., 1975). The induced current may also be responsible for the spread of biogenic particles derived from the abrasion of the reef complex during wave episodes of high energy. The resulting flux is expected to impact residence time along the inner reef. On day 2 of the experiment, during spring tide, velocity averaged over one tidal cycle was higher than 0.4 m s1 (Fig. 7). If we extrapolate this value along the barrier reef (approximate length: 100 km), the total flux of water above the reef is estimated to be 4 104 m3 s1. This value corresponds to one third of the incoming tide-related fluxes from the South and the inlets during spring tide. From days 3 to 7, averaged velocity ranged from 0.15 to 0.25 m s1, and the flux was between 1.5 and 2.5 104 m3 s1, i.e. around 20% of the incoming flux from the South

and through the inlets between spring and neap tide. These estimations indicate that residence time must be significantly changed in this case. Simulations taking into account the water inputs through the barrier reef were performed. The velocity given by Eq. (2) was added at each grid points of the barrier reef to the currents calculated by the hydrodynamic model, after each time step. The difference between the local e-flushing time without taking account wave breaking and the one calculated taking into account wave breaking is mapped on Fig. 8. These preliminary results show that when waves break on the reef, residence time changes significantly along the barrier reef in the northern inner reef where water renewal is dramatically constrained by narrow passages. Along the southern part of the reef the decrease is less important since water is mainly renewed from the water entrance to the South and above all because the e-flushing time was very low. It decreases from 4– 5 days without wave breaking to 1 day with wave breaking in the behind Aboré reef area. In contrast, between the inner reef and the outer lagoon, residence time increases by 3–5 days. Along the coast and in the bays, the change in residence time is very low and almost uniformly distributed and the spatial patterns of residence time are not altered. The gyre with long residence time which extends to the West of the Woodin channel is probably slightly enlarged. The flushing lag (not shown) drops off in the back-reef areas far from the passages. The full analysis of the results is still underway and the complete results will be submitted for publication. The paper by Bonneton et al. (2007) quantified the global mass and momentum fluxes over the reef in case of weak wind and high oceanic swell propagating nearly normally to the reef as encountered during the first week of measurement. The applicability of a wave pumping model by Callaghan et al. (2006) to our dataset and a finer analysis of the wave transformation processes over the reef (Bonneton et al., 2008) are being determined. 3.4. Wind waves Apart from river inputs which are very episodic, the main mechanism responsible for turbidity is the combined effect of waves and currents in resuspension of particles (Clavier et al., 1995). Under

WT

North

20

East

10

South

H sF (m)

West North

0

1

WWIII ADV

0. 5

Wind Speed (m s -1)

S. Ouillon et al. / Marine Pollution Bulletin 61 (2010) 269–296 Wind direction (°N)

278

T02F (s)

0 3 0

Fig. 9. Wind forcing and comparison between measured (after swell filtering) and simulated significant wave height and wave period from WaveWatch3, station WT (see location Fig. 10), August 8–21, 2006.

usual wind conditions, resuspension is particularly significant where the water depth is less than approximately 20 m similar to the situation in the Great Barrier Reef area (Larcombe et al., 1995, 2001). At greater depths, the effect of waves becomes negligible, except during storms. Studying particulate transport thus also requires studying generation and propagation of wind waves in the lagoon, since the locally generated wind waves are the main component of the sea state. We therefore both measured and modeled waves. The NOAA wave model WaveWatch3 (Tolman 1991, 1999) was used to simulate wave distribution in the lagoon, and wave parameters were measured in situ by a wave and tide recorder (WTR9 Aanderaa) and by an acoustic Doppler velocimeter (ADV Sontek). More details on wave modeling are given in Jouon et al. (2009). Model validation in the SLNC was necessary because the lagoon is a semi-enclosed and fetch-limited basin. Furthermore, a validated wave model is required to simulate the extreme wind seas during storms. Special care was taken in the validation step as, on one hand the wave model did not simulate the propagation of low-frequency (63 lm), except for in the near-shore areas (Clavier et al., 1995), has traditionally been measured in the particles 80%) along the reef, as defined by the classification of Maxwell (1968). Canyons are dominantly filled with marine muds. Content of the residual fraction varies inversely with carbonate contents indicating the terrigenous origin of these particles. Close to river mouths, they vary from 77% (B03, la Coulée R. mouth), 62% (P04, Pirogues R. mouth), 54% (D36, Dumbéa Bay), 33% (N12, Sainte-Marie Bay), to 10% or less in the vicinity of inlets (Magand, 1998). Resuspension occurs within the lagoon during each tidal cycle. Clavier et al. (1995) show that at each of five stations typical of the three bottom types, more than 80% of the total sedimentation was linked to deposition of resuspended benthic material. This feature was particularly conspicuous for their two muddy substrate stations where a 90% rate was determined. The seafloor geology appears to be well correlated to the sediment transport pattern, as shown by the agreement of erosion and deposition areas determined from numerical simulation (distinguished by the local bed shear stress) and the map of mud content in the superficial sediment in Douillet et al. (2001). Sediment transport and resuspen-

sion have led to a strongly sediment-partitioned shelf, with modern mud-rich sediments almost exclusively restricted to the inner and inner-middle shelf, to northward-facing embayments and in the lee of headlands. Hardground indicates an erosional environment, and mud deposits evidently suggest depositional environments. Mineralogy of surface sediments determined by X-ray diffractometry shows larger differences between the bays than does the mineralogy of suspended particulate matter (SPM). Among terrigenous sediments, quartz and goethite dominate at the Pirogues River mouth, talc and goethite at the Coulée River mouth, antigorite and quartz at the Dumbéa River mouth. Moreover, the Coulée River also discharges kaolinite which was undetectable in the Pirogues and Dumbéa Rivers (Magand, 1998). Carbonated compounds (calcite, magnesium calcite and aragonite) are always present and in increasing quantities as one moves away from the bay heads. The sedimentological maps (Dugas and Debenay, 1982) and the geochemical analyses (Magand, 1999) reveal the mixture of fine limestone sands and of lutites of siliceous origin (terrigenous) in the sand-muddy deposits in gray bottoms, canyons and at the vicinity of mouths. It is a delicate exercise to compare the particle inputs resulting from land erosion to sedimentation within the lagoon. Indeed, most of the terrigenous input is trapped in deltas and mangroves, and only a minute part reaches the lagoon (Clavier et al., 1995). The terrigenous influence is lower in Dumbéa Bay than in Boulari Bay and near the Pirogues River mouth for two distinct reasons. Firstly, the early trapping of material in alluvial plain slows down particulate transit between the Dumbéa catchment area and the middle of bay. Baltzer and Trescases (1971a) estimated that three quarters of the particles brought by the Dumbéa River settle in the delta and that another significant part is also retained in mangroves. Secondly, in addition to a lower terrigenous input, Dumbéa Bay also shelters fringing reefs and alive or dead corals on its bottom (Launay, 1972) and these constitute an important source of carbonates as compared to the Coulée and Pirogues River mouths. Moreover, the Coulée and Pirogues Rivers have particularly striped basins as a consequence of scrubbing of lateritic soils. The absence of a deltaic plain and weak mangrove coverage at the mouth further supports the hypothesis of increase discharge of terrigenous material in the lagoon. The higher percentage of sediments of terrigenous origin in Boulari Bay has been confirmed by geochemical analyses (Magand, 1999). The fringing reefs in the west of Boulari Bay (around Tina and Sainte-Marie Bay) probably explain the higher presence of biogenic particles in this zone.

282


Sainte-Marie Bay shelters a fringing reef, receives terrigenous inputs from the Coulée River (Fernandez et al., 2006), and collects domestic and urban waste waters. It is thus a particularly complex zone from the sedimentological point of view. It should be also noted that the content of residual fraction of sediments, like that of total carbon, decreases with depth in sedimentary cores. This indicates that terrigenous particles deposited on the surface are resuspended during high-energy events like storms and are redistributed by currents. Thus the distribution of freshly deposited particles is spread over the long term after a succession of transient episodes of resuspension/transport and deposit. 4.2. Suspended matter concentration and turbidity: spatial patterns and variability 4.2.1. Nature and characteristics of the suspended matter The input of elements into the lagoon occurs in three forms: dissolved, colloidal and particulate, each of which corresponds to distinct biological and physico-chemical reactivities. The oceanographic community has adopted an arbitrary threshold at 0.45 lm to distinguish dissolved and particulate elements by filtration (Stumm and Morgan, 1981) and the colloidal state, a state of transition, relates to particles whose size lies between 1 nm and 1 lm (Vold and Vold, 1983). Suspended particles have high capacities of chemical exchange and thus are vectors of metals, chemical and organic pollutants (Ouillon, 1998). In New Caledonia, the study of elementary fluxes has focused on the fate of suspended particles and of dissolved and particulate metals. Suspended particles in coastal waters include non cohesive mineral particles, fragments, plankton and aggregates. Plankton temporal variability is described in a companion paper (Le Borgne et al., 2010). Aggregates (from several micrometer to several mm) consist of inorganic and organic matter, marine snow, fecal pellets, plankton, mineral particles, microbial communities, particulate organic matter originating from dissolved organic matter like Transparent Exopolymer Particles (TEP; Azetsu-Scott, 2004) and other Extracellular Polymeric Substances (EPS) or a mixing of these ingredients with water (Droppo, 2001). Microbial secretions of EPS form a framework and provide structural stability of the aggregates. Physico-chemical processes and biological activities continuously modify aggregate size, stability, density, porosity, and biochemical processes (Droppo, 2004) with their buoyancy varying as a function of the particles contained within the aggregate. Moreover, these suspended particles react quasi-instantaneously to the forcing applied to the studied system. Semi-quantitative mineralogical analysis by X-ray diffractometry showed the joint presence of terrigenous and carbonaceous detritic minerals in SPM collected in May–June 1998 (Magand, 1998). The most abundant detritic terrigenous minerals were, in decreasing order of importance, quartz and two magnesium clays, antigorite and talc. SPM in bays also contained nontronite (ferralitic clay; Dumbéa, Boulari, Pirogues) and goethite (iron hydroxides; Dumbéa, Boulari). The main carbonaceous minerals were calcite, magnesium calcite and aragonite. The terrigenous fraction was more abundant in the heads of bays, while the carbonaceous fraction dominated in the lagoon (Magand, 1998). Scanning electronic microscope (SEM) allowed the identification of three main phytoplankton groups in the SPM from the lagoon: diatoms are dominant in the bays except Dumbéa; coccolithophorids are abundant in Dumbéa Bay, close to the reefs, canyons and inlets; dinoflagellates are somewhat abundant, with slightly higher abundances in Sainte-Marie Bay than elsewhere. Pelagic diatoms dominate in biomass, benthic diatoms in number. The presence of benthic diatoms (in particular Diploneis sp. and Amphipora augustata) in sediment traps results from the resuspen-

sion of deposited surface materials. River inputs of dissolved nutrients, in particular in silica, probably explain the abundance of diatoms in coastal areas compared to further offshore. Abundant coccoliths are characteristic of the near ocean (Jacquet et al., 2006). The organomineral particles, observed by SEM, resulted from a mixture of two distinct populations (Magand, 1998): abundant large aggregates of several hundred microns containing carbonated detrital minerals (mainly coral fragments composed of fibrous aragonitic structures), and smaller, less abundant aggregates of about 10 lm, containing finer terrigenous detrital minerals (smoothed or polished siliceous particles). These smaller particles were comparable with those described at the Dumbéa River mouth by Baltzer and Trescases (1971b). Rivers export sedimentary material composed of products from scrubbing and hydrolysis of peridotitic cover (laterites and saprolites) of the catchment areas as well as the sedimentary rocks from the piedmont or alluvial plain. The origin of material explains the quasi-total absence of carbonated compounds (